Door-to-door transit accessibility using Pareto optimal range queries

Abstract

Public transit is a backbone for well-functioning cities, forming a complicated system of interconnecting lines each with their own frequency. Defining accessibility for public transit is just as complicated, as travel times can change every minute depending on location and departure time. With Pareto optimal journeys it is possible to look beyond the earliest arrival times and also optimize for the shortest travel time, as travellers base their departure time on the start time given by their smartphone app, especially when service frequencies are low. By querying for all Pareto optimal journeys in a time range it becomes possible to get a grasp of what passengers see as their choice set when it comes to transit route choice. Based on the averages of the Pareto optimal journeys it should become possible to calculate more realistic skim matrices for traffic analysis zones, including reliability factors such as frequencies and the number of transfers. In this study we calculate Pareto optimal journeys in the area in and around Amsterdam, looking at how travel times are distributed and what factors impact them. Public transportation is an important travel mode that keeps cities liveable. Determining travel times for public transport alternatives is a more difficult task than for pedestrians, bicyclists and even cars. This has multiple reasons, starting with the fact that public transportation always involves another modality such as walking or cycling, meaning that public transportation accessibility is heavily dependant on the distance from the origin to the nearest location to board a transit vehicle and the distance between the destination and the nearest location to disembark transit. A main reason that analysis of public transportation is difficult is that we are dealing with a time-dependent network, as transit is an intricate system of buses, trains, trams, metros that drive in frequencies that change depending on the time-of-day and are affected by external factors such as traffic and weather. To compensate for these external factors, timetables often include extra time in the travel time and transfer times, so passengers will still be able to make transfers in case of minor delays and the vehicle will just wait in locations where possible and convenient. Finally there are situations where there is trade-off between the access and egress distance and the total travel time. For example, a bus stop next Abstract Public transit is a backbone for well-functioning cities, forming a complicated system of interconnecting lines each with their own frequency. Defining accessibility for public transit is just as complicated, as travel times can change every minute depending on location and departure time. With Pareto optimal journeys it is possible to look beyond the earliest arrival times and also optimize for the shortest travel time, as travellers base their departure time on the start time given by their smartphone app, especially when service frequencies are low. By querying for all Pareto optimal journeys in a time range it becomes possible to get a grasp of what passengers see as their choice set when it comes to transit route choice. Based on the averages of the Pareto optimal journeys it should become possible to calculate more realistic skim matrices for traffic analysis zones, including reliability factors such as frequencies and the number of transfers. In this study we calculate Pareto optimal journeys in the area in and around Amsterdam, looking at how travel times are distributed and what factors impact them. Public transportation is an important travel mode that keeps cities liveable. Determining travel times for public transport alternatives is a more difficult task than for pedestrians, bicyclists and even cars. This has multiple reasons, starting with the fact that public transportation always involves another modality such as walking or cycling, meaning that public transportation accessibility is heavily dependant on the distance from the origin to the nearest location to board a transit vehicle and the distance between the destination and the nearest location to disembark transit. A main reason that analysis of public transportation is difficult is that we are dealing with a time-dependent network, as transit is an intricate system of buses, trains, trams, metros that drive in frequencies that change depending on the time-of-day and are affected by external factors such as traffic and weather. To compensate for these external factors, timetables often include extra time in the travel time and transfer times, so passengers will still be able to make transfers in case of minor delays and the vehicle will just wait in locations where possible and convenient. Finally there are situations where there is trade-off between the access and egress distance and the total travel time. For example, a bus stop next